The 2026-10-27 YZ Index Smoke Quick Test covered 13 models, with Claude Opus 4.7 ranking first for the day at 87.7 points. Smoke is a daily 10-question quick test, suitable for observing short-term signals and not equivalent to conclusions from the Full weekly leaderboard.
This Smoke evaluation covers only two main leaderboard dimensions: Code Execution and Material Constraints. The main leaderboard formula is 0.55 × Code Execution + 0.45 × Material Constraints. Because the daily sample size is small, single-day scores are better used as monitoring signals rather than long-term conclusions about model capability.
Daily Rankings
| Rank | Model | Main Leaderboard | Code Execution | Material Constraints | Integrity |
|---|---|---|---|---|---|
| #1 | Claude Opus 4.7 | 87.7 | 99.8 | 72.9 | pass |
| #2 | GPT-5.5 | 78.49 | 100 | 52.2 | pass |
| #3 | Gemini 3.1 Pro | 77.34 | 75 | 80.2 | pass |
| #4 | GPT-6 Astra | 76.65 | 100 | 48.1 | pass |
| #5 | Claude Sonnet 4.6 | 76.54 | 99.8 | 48.1 | pass |
| #6 | GPT-6 Luna | 72.46 | 100 | 38.8 | pass |
| #7 | Grok 4 | 72.28 | 94.6 | 45 | pass |
| #8 | GPT-6.1 Sol | 62.9 | 75 | 48.1 | pass |
| #9 | GPT-6 Sol | 61.91 | 75 | 45.9 | pass |
| #10 | GPT-o3 | 60.31 | 50 | 72.9 | pass |
| #11 | DeepSeek V4 Pro | 56.84 | 50 | 65.2 | pass |
| #12 | Gemini 2.5 Pro | 55 | 50 | 61.1 | pass |
| #13 | Doubao Pro | 45.34 | 25 | 70.2 | pass |
Data Interpretation
In today's YZ Index Smoke Quick Test, Claude Opus 4.7 ranked first with a main leaderboard score of 87.7. Its Code Execution at 99.8 and Material Constraints at 72.9 form a high-level combination, indicating that both contribute in a balanced way to the formula result of 0.55 × Code Execution + 0.45 × Material Constraints. GPT-5.5 follows closely with a main leaderboard score of 78.49; its Code Execution is 100 but Material Constraints is only 52.2, reflecting a structure dominated by Code Execution. Gemini 3.1 Pro, with a main leaderboard score of 77.34, shows a relatively balanced combination of Code Execution at 75 and Material Constraints at 80.2. GPT-6 Astra, with a main leaderboard score of 76.65, likewise relies on Code Execution at 100 while Material Constraints is 48.1.
Compared with the previous run under the same methodology, GPT-5.5's main leaderboard score rose 46.3 points, mainly from Code Execution rising 92.6 points while Material Constraints fell 10.2 points; Claude Opus 4.7's main leaderboard score rose 36.7 points, accompanied by a 70-point rise in Code Execution; Gemini 3.1 Pro's main leaderboard score rose 34.3 points, with Code Execution up 50 points and Material Constraints up 15 points. GPT-o3's main leaderboard score rose 24.9 points, corresponding to a 50-point rise in Code Execution, while GPT-6.1 Sol's main leaderboard score fell 23.2 points, accompanied by a 25-point drop in Code Execution and a 20.9-point drop in Material Constraints. These single-day changes may stem from question sampling fluctuations or may reflect real changes in model performance and require follow-up runs for confirmation.
The Smoke Quick Test is a small-sample, single-day signal. There are currently no anomaly signals, and all models have Integrity pass. Interpretation is limited to analysis of today's data structure and does not make long-term inferences.
Key Changes
- GPT-5.5: Main leaderboard up 46.3 points, Code Execution +92.6 points, Material Constraints -10.2 points
- Claude Opus 4.7: Main leaderboard up 36.7 points, Code Execution +70 points
- Gemini 3.1 Pro: Main leaderboard up 34.3 points, Code Execution +50 points, Material Constraints +15 points
- GPT-o3: Main leaderboard up 24.9 points, Code Execution +50 points, Material Constraints -5.9 points
- GPT-6.1 Sol: Main leaderboard down 23.2 points, Code Execution -25 points, Material Constraints -20.9 points
Signals to Watch
- No anomaly signals were retained for publication in this run.
When reading this type of Smoke brief, the focus should be on two questions: first, whether a model exposes the same type of weakness on consecutive days; second, whether its Integrity rating moves from pass to warn or fail. Large single-day changes in execution or constraint scores may come from question sampling or may be early signals of real degradation, requiring follow-up runs for confirmation.
Data source: YZ Index | Run #398 | View raw data
© 2026 Winzheng.com 赢政天下 | 转载请注明来源并附原文链接